Divergent Square Averages

We answer a question of J. Bourgain. We show that the sequence n^2 is L^1-universally bad.